Use torque as the job scheduler. To use torque from yum, you need to add EPEL repository. Install and configure as follows.
$ su -
$ rpm -Uvh http://ftp.riken.jp/Linux/fedora/epel/6/x86_64/epel-release-6-8.noarch.rpm
$ yum install --enablerepo=epel torque-server torque-client torque-mom torque-scheduler
$ /usr/sbin/create-munge-key
$ hostname > /etc/torque/server_name
$ pbs_server -t create
After this, edit qmgr.txt as follows:
create queue L0 queue_type = execution
set queue L0 enabled = true
set queue L0 started = true
set server default_queue = L0
set server scheduling = true
set queue L0 resources_max.ncpus = 12 #← For 12 cores
set queue L0 resources_max.nodes = 1
Then, operate as follows.
$ service trqauthd start
$ qmgr < qmgr.txt
$ echo `hostname` "np=12 num_node_boards=1" > /var/lib/torque/server_priv/nodes
$ sed -i.bak s/localhost/`hostname`/g /var/lib/torque/mom_priv/config
$ echo "nodes=1" > /var/lib/torque/mom_priv/mom.layout
$ service pbs_server restart
$ service pbs_mom restart
$ service pbs_sched restart
$ chkconfig pbs_mom on
$ chkconfig pbs_sched on
$ chkconfig pbs_server on
$ chkconfig munge on
$ chkconfig trqauthd on
Recommended Posts